/ EN
2026-09-16 00:00 Papers Foundations & Methods Translated from EN

iMINDBench Sets a Shared Cross-Institution Test for Intracranial EEG Decoding

Summary Intracranial EEG (iEEG), recorded by electrodes implanted inside the brain, is widely regarded as an ideal signal for decoding intent, but differing datasets and preprocessing pipelines make it hard to tell whether models are actually improving. A research team built iMINDBench, a benchmark that brings together naturalistic movie-watching data from three institutions, 15 decoding tasks, standardized preprocessing and fixed evaluation splits. Pretrained systems generally beat baselines within their own preprocessing pipeline, but classic spectral baselines remained competitive on other institutions' data. Scaling up supervised data from other subjects or institutions to 25 times the volume yielded only limited, task-dependent gains over training on data from the same session.
Why it matters Whether data from one institution carries over to another is a question intracranial EEG decoding cannot sidestep on the way to general-purpose models. By turning preprocessing differences into separate comparison tracks, the benchmark draws a line between a better model and better-cleaned data. That 25 times more external data brings only limited gains suggests data volume is not the current bottleneck; stable cross-subject, cross-institution decoding is.

BCIwiki (bciwiki.com) — Researchers have released iMINDBench, a multi-institution benchmark for intracranial EEG (iEEG) neural decoding that puts three naturalistic movie-watching datasets, 15 decoding tasks, standardized preprocessing tracks and fixed evaluation splits into one evaluation framework. The preprint was posted to arXiv on September 16, 2026, and has not been peer reviewed.

Intracranial EEG records electrical activity directly from electrodes inside the human brain, making it an attractive modality for neural decoding. But the team notes that progress in the field is hard to measure reliably: datasets are task- or institution-specific, limiting evidence of generalization across tasks and recording environments, and preprocessing choices strongly influence performance, making model improvements difficult to distinguish from preprocessing gains.

Evaluated on iMINDBench, the pretrained systems tested generally outperformed baselines within their respective preprocessing tracks, while strong spectral baselines remained competitive across institutional datasets.

A scaling study produced a blunter result: adding up to 25 times more supervised data from other subjects or institutions yielded only small or task-dependent gains over within-session training. The team concludes that iEEG models need to improve on strong preprocessing baselines and make more effective use of data across subjects and institutions.

The preprint runs 26 pages with 13 figures. Its authors are Geeling Chau, Saba Hashemi, Yonghyeon Gwon, Eshani Patel, Jan DeWitt, Christopher Wang, Andrii Zahorodnii, Sabera J Talukder, Danny Dongyeop Han, Chun Kee Chung, Maryam M Shanechi and Yisong Yue.

Compiled by BCIwiki from public sources

Sources · 1
arxiv.org 2026-09-16
Read original ↗
Suggest a correction Revisions · none / EN
© 2026 BCIwiki.com Digest Topics Tips Subscribe Revisions About